The Two Surfaces of Google Duplex
At this year’s Google I/O Developer Conference, Google Duplex, an AI system caught my eye. Google Assistant was able to make calls on behalf of its user and do simple tasks such as booking an appointment or making a reservation at a restaurant. The system uses a Recurrent Neural Network while interacting with humans. The network takes note of three components: the text recognized by the automated speech recognition technology (ASR), features of the audio, and the context of discussion [1]. For instance, to make a reservation, the assistant understands the queries posed by the staff, takes into account their accent and also focuses on the various parameters of the conversation – before preparing a response. Like most individuals, I was amazed to see this interaction between a human and a machine – because if not told, one could not make out, which of them was the machine. The AI system was able to intonate and precisely pause to maintain a natural style of conversation. The two discussions, led by the system, ranged from being easy to involving complexities; however, the AI was able to competently handle both conversations and do the work for its users. Pretty impressive isn’t it?
In the near future, this application can be used to create positive outcomes for various individuals in society. To elaborate, any amount of time saved by an individual can be channelized into something more productive that is outside the ambit of what the professional world expects you to do. Leisure and recreational activities help in the acquisition of new skills and development of the human personality that is essential to survive in status quo. By using Google Duplex, the processes of learning can be made more interactive for children and adults with disabilities, as this bot is unfamiliar with the ideas of fatigue. Duplex will also facilitate in eliminating social constructs like discrimination on the basis of race, community, and gender etc. This will make processes like confession recordings and interrogation techniques simpler and more efficient. I feel any benefit to the state structure trickles down to all units of society and therefore Google Duplex is a technological leap towards a better society.
However, this bot is not able to make general conversation and is an endeavor towards a mechanical experience of the real life. I feel there is a prominent distinction that exists between a technologically progressive world and a world that results in human laziness and languor. The introductory comment of Duplex being an assistant is not just monotonous but also poses an inherent contradiction. Why does Duplex have to sound so bumbling and cheery when it is actually just a robot? The Google IA blog even states that natural disfluencies such as “Mm-hmm” are designed into the assistant to make this system sound “more familiar and natural” [2]. A subsequent question can be to what degree can Duplex imitate a human being? The underlying emotions and feelings are pretty complex while considering the comprehensive ability of a machine. Our speech does not only consist of words but many pauses and conjunctions, which can easily be misconstrued. Another question taking rounds is, what will happen in a situation where a Google Duplex is communicating with another over a telephonic conversation? Which way will the conversation go, if an enterprise such as a call center takes up this system? Will structural unemployment take place [3]?
Lastly, I feel the issue being disregarded is the human psyche. We live in a consumer conscious environment, where every person is largely specific about his or her demands. As a rational person, will I be able to trust a machine to make an appropriate appointment taking care of all my demands? In this trade-off, will I prefer to feed in my details to Google Duplex or physically make an appointment myself? In a world where it is hard to decipher if the news is fake or if a particular photo can be trusted, is it beneficial to know that you can’t even believe your ears [4]?
Thus, Google Duplex can be a blessing or an evil we’ve created and accepted for ourselves. Now, it is for us to choose which side of the coin we prefer.
References:
[1] – https://www.youtube.com/watch?v=FEhLW1cX4c0
[2] – https://ai.googleblog.com/2018/05/duplex-ai-system-for-natural-conversation.html
[3] – https://www.investopedia.com/news/google-duplex-ai-ready-replace-call-centers/
[4] – https://www.cnet.com/news/google-duplex-assistant-bot-deception-scary-ethics-question/
Users who have LIKED this post:
3 comments on “The Two Surfaces of Google Duplex”
Comments are closed.
This is a really interesting article! Particularly because Duplex has a wide potential range of uses but right now its usage is being (perhaps unnecessarily?) limited.
Currently, Duplex is being marketed as an assistant, but an AI that can mimic human voice and sustain a perfectly human conversation could have so many other uses – such as therapy for mental health. What is now seen as a solution to very trivial problems such as handling restaurant reservations could potentially become a solution to the very pressing issues that we are facing now.
And with that, I wonder if such developments will isolate us from one another as human beings — why bother making friends if there exists AI systems that will actually listen to our troubles, comfort us when we’re sad, and rejoice at our success, without all the drama that comes with friendship? We’ve seen how many of the lonely turn to the internet and chatbots for temporal company and respite, and the fact is that putting a human voice to an AI system makes the companionship seem so much more real, so much more tangible.
Of course, you’re right about there having two sides of the coin — while potentially socially isolating, such developments could also really help the world, say, if such systems could detect early symptoms of mental illness and deliver personalised responses to help people cope, the world would be a much better place.
I suppose this then brings the discussion to the sort of policy that needs to be put in place to address the issue. Currently, it seems sufficient that the AI system should identify itself as one before the conversation to curb the ‘unethicalness’ of the situation. However, when the applications of such developments are to be expanded, perhaps we need to ask ourselves if identification is sufficient, or if other safeguards need to be put in place.
I really liked your stance and how you’ve shown both the sides. Its interesting what it can do to people who decide to use it. For instance, by outsourcing conversations we’d rather not have to bots in order to get more things done, we may chip away at our own humanity.
As David Ryan Polgar said: “In our pursuit to really communicate with more people than is humanly possible, we’re potentially deleting the very value of our communication.”
Wow, what a great article.
When I first heard about Duplex I was also left dumbfounded by its ability to take part in conversation, but I had never put any thought into “the other side of the coin”. There are so many things that come to mind when I think all of the possible outcomes that this technology’s uses could have on society. This technology could cause mass losses in employment and major trust issues in people that won’t know if what they hear is real.
As you stated in your article, why instead of bothering to tell Duplex to call the restaurant wouldn’t you just call it yourself? With the increasing amount of “screen time” and the use of cellphones social anxiety has started to take over. People would rather use Duplex than call themselves to avoid engaging as much as possible. This is because we keep creating technologies that reduce our interactions with people with everything from cashier-less stores to driverless cars. Duplex just sets up another barrier between people.
This also opens up the concern that at one point, what if people started to treat Duplex as an ideal “friend” or “personal companion”. People could be tricked by the “realness” in the voice and its ability to cater to their needs into replacing their real friends and relationships with “virtual friends”.
As you also stated in your article, this technology is a two sided coin. As a society we have to take a step back and really think on how we want to use this technology, using this technology could really make a positive impact on society, but “with great power comes great responsibility”.
Users who have LIKED this comment: